2.1 Experimental animals and feeding
Glass eels of A. japonica and A. marmorata were caught from eastern Taiwan (A. japonica from the Yilan River, 24.7163 °N, 121.8348 °E, and A. marmorata from Xiugulan River, 23.4612 °N, 121.5008 °E). Eel sampling was approved by the Fishery Agency, Council of Agriculture, Executive Yuan, Taiwan. The specimens were transported at low temperatures through live fish bags filled with oxygen. The health condition of the eels was checked upon arrival at the laboratory located at the Institute of Fisheries Science of National Taiwan University, Taipei. Individuals in good condition were disinfected with 2.5 ppm of potassium permanganate (KMnO4) solution for 10 min to avoid pathogen contamination of the experimental system. After sterilization, the eels were kept in five sets of indoor RAS systems with five tanks (30 × 30 × 45 cm) for each set and maintained in freshwater for three days before feeding. Photoperiods were set at 12 h light (7:00–19:00) and 12 h dark.
The initial body weight and total length of A. japonica and A. marmorata (20 A. japonica for each tank; 30 A. marmorata for each tank in triplicates) were measured before experiment started (56.7 ± 2.0 mm, 0.14 ± 0.01 g for A. japonica; 51.04 ± 2.1 mm, 0.15 ± 0.02 g for A. marmorata). An LED (EVERLIGHT Electronics Co., Ltd., Taiwan) was used as the light source to control the background spectra for the experiment. Each set of RAS included five tanks (30 L water/tank), each exposed to either white light, red light (622 nm), green light (517 nm), or blue light (467 nm) under 100 lx light intensity with photoperiod 12 hours light and 12 hours dakr, or dark (<5 lx). Each RAS tank was covered by a black board to avoid any light influence from neighbouring tanks or the environment (Fig. 1). The water temperature and pH were controlled between and , respectively, with a water exchange rate of 20 L/day for each RAS; oxygen was dissolved to near saturation by aeration. Fish were fed with blood worms (Chironomus dorsalis larvae), which was often used as glass eel feed, at an amount of 10% of their body weight twice a day for a total of 12 weeks. The remaining feed was removed from the tanks an hour after feeding. The experiment was performed in accordance with the recommendations from the Institutional Animal Care and Use Committee for the care of animals used for experimental or other scientific purposes (approval number ‘NTU-110-EL-00009’).
2.2 Sample Collection and Analyses
The total length (to the nearest 0.1 mm) and body weight (to the nearest 1 mg) were measured every two weeks. The percentage weight gain, condition factor, specific growth rate, and the survival rate in each group were calculated as follows:
Three fish from each tank were randomly selected and sacrificed to obtain head kidney tissues. Three head kidney tissues from the same tank were pooled together and stored in an RNA protecting reagent at −80 before extracting total RNA using an RNA kit (Bioman Scientific Co. Ltd., Taiwan) for next generation sequencing (NGS). Three other eels from each tank were randomly selected and sacrificed to obtain the head kidney, and pooled together for real-time PCR of immune-related genes.
2.3 NGS
The whole genome of A. japonica was successfully assembled in our previous study (http://molas.iis.sinica.edu.tw/jpeel/) (Hsu et al., 2015), and was used as a template to annotate the transcriptome data of A. japonica and A. marmorata. The head kidney samples of the white light, red light, green light, and blue light groups were preserved at −80 and subjected to NGS for transcriptome analysis of the head kidney to identify the immune-related genes in each group. The raw RNA-seq data were filtered using the TrimGalore program (Babraham Bioinformatics, Cambridge, UK) to discard adaptors and low-quality reads (Q < 13). Low-complexity reads (repeat sequences) were then removed using the PRINSEQ program (ver. 0.20.4). Finally, general read properties were generated using the FastQC program (Babraham Bioinformatics, Cambridge, UK). The MAKER2 (https://www.yandell-lab.org/ software/maker.html) pipeline was chosen for gene prediction and the gene transfer format was generated. The input datasets included RNA-Seq data, PacBio Iso-Seq data, the newly established A. japonica genome, available protein sequences of A. japonica, all teleosts from the uniport database, and the pre-existing zebrafish (Danio rerio) gene model. The predicted open reading frames were annotated with homologs in the NCBI nr database (April 2019) using GhostX (Suzuki et al., 2014). Protein sequence features, including signal peptides, transmembrane domains, and domains described in the Pfam database, were detected by SignalP (Petersen et al., 2011), TMHMM (Krogh et al., 2001), and HMMER (3.1b2) (Mistry et al., 2013). Gene Ontology (GO) annotation was performed for genes with detectable Pfam domains according to Pfam2GO. Furthermore, the protein-coding genes were mapped to the canonical pathway database KEGG pathway using the KEGG Automatic Annotation Server (Moriya et al., 2007), setting the parameters for mapping only to the related prokaryotic database in single-directional best hit (SBH) mode; the signal of an innate immune gene can be detected using this website.
The NGS results were further uploaded to the website (http://molas.iis.sinica.edu.tw/jpeel2018/index.php) set up by our laboratory and Academia Sinica for assembly and analysis. This web database is established on the LAMP system architecture (Ubuntu 14.04, Apache 2.04, PHP 5.1 and MySQL 8.0) with the Bootstrap 3 CSS framework (http://getbootstrap.com/), jQuery1.11.1, and jQuery Validation v1.17 to provide an intuitive user experience. The entire system runs in a virtual machine on the cloud infrastructure of the Institute of Information Science, Academia Sinica, Taiwan. The analysis was performed using scripts written in R (3.4.2). According to the assembly with gene transfer format, raw reads generated from RNA-seq were estimated for the expression profiles via an intuitive graphical interface in Docexpress (https://hub.docker.com/r/lsbnb/docexpress _fastqc) with a built-in process of Hisat2 → StringTie → Ballgown. The expression level profiles, in the form of fragments per kilobase of transcript per million mapped reads values, were then submitted to Multi-Omics onLine Analysis System (http://molas.iis.sinica.edu.tw/jpeel2018/index.php) for both eel species.
After this analysis, four innate immune-related genes, viz., lysozyme (LZM), superoxide dismutase (SOD), peroxidase (POD), and interleukin-6 (IL-6), which could be detected by NGS in both eels, were selected for real-time PCR (Table 2).
2.4 Real-time PCR
Specific candidate genes were selected for real-time PCR based on the results of the NGS analysis. TRIzol reagent (Bioman Scientific Co. Ltd) was used to extract total RNA, and the purity was quantified by spectrophotometry (Medclub Scientific Co. Ltd). Reverse transcription was performed to synthesize complementary DNA (cDNA) for real-time PCR (Bio-Rad). Four immune-related genes, namely, superoxide dismutase (SOD), lysozyme (LZM), peroxidase (POD), and interleukin-6 (IL-6) were selected as the target genes for real-time PCR, and acidic ribosomal protein (ARP) was used as the reference gene. The primers used for real-time PCR are listed in Table 1.
2.5 Statistical analysis
All data were analysed by one-way analysis of variance (IBM SPSS Statistics 24.0) to determine the effects of different spectra. Statistical significance was set at P < 0.05. A significant effect was followed up with the least significant difference test to compare the means.